PR Review Summary

Status: Request Changes

🪜 Strategic-Fit Decision

Per §9 Strategic-Fit Step-Back:

• Decision: Request Changes
• Rationale: The three-way split is the right architectural response and is salvageable in place. The exact head still leaves producer/consumer identity mismatches, an obsolete Mission Control identity document, an unsatisfied e2e-green close-target AC, and unrecorded runtime coupling; these are delivered-scope defects, not reasons to discard the relocation.

The import/URL split is thoughtful, the portal card follows the route it historically booted, and the guard correction is evidence-led. The remaining issues are concentrated in runtime identity and public authority/evidence.

🧭 Patch-Blind Premise Snapshot

• Inputs Read Before Patch: #16322 body plus all four current decision comments; 52-file changed list; current dockdemo, examples/dashboard/{base,dock,crossWindowWitness}, AgentOS childapp, portal-catalog, and app-work core contracts; structural-preflight destination receipt 5380690099; prior docking summary summary_54156254-a1a8-40b3-ba22-86e7d2a1bf81 (session 54156254-a1a8-40b3-ba22-86e7d2a1bf81) recording Demo A’s deterministic/whitebox role.
• Expected Solution Shape: Demo A/B and their scripts become self-contained examples/dashboard/ apps; Mission Control remains only if its production ownership is explicit; widget stays with product infrastructure. The move must preserve URL, App-Worker identity, themes, portal metadata, and full whitebox behavior; it must not claim independence while loading the product shell by hidden string dependency, and test isolation must compare the same observable contract rather than only failure names.
• Patch Verdict: Partially matches. Files and ESM imports split coherently, but Neural Link fallbacks still name the retired app, Mission Control’s README/HTML still describe the deleted demo app, five e2e failures leave AC-2 open, and the product-theme dependency survives outside the import census.
• Premise Coherence: The relocation aligns with the Body/product boundary and verify-before-assert. Calling different assertion failures “byte-for-byte parity,” and leaving the ticket’s key runtime surfaces without a Contract Ledger, breaks that same evidence discipline.

🕸️ Context & Graph Linking

• Target Epic / Issue ID: Resolves #16322
• Related Graph Nodes: #15614, #16315, #16316, #17241, #17504, #17539, #17541, #17555
• Origin Session ID: f47f948b-743b-4c11-84a8-fa60a567a148

🔬 Depth Floor

Challenge: Exact-head source contradicts three delivery claims:

1. The three new apps declare Neo.examples.dashboard.choreography, Neo.examples.dashboard.crossWindow, and AgentOSMissionControl, while every moved dock e2e still passes AgentOSDockDemo as its explicit connectToApp() fallback. The current worker-id fast path can hide that stale fallback; fixtures.mjs:169-180 documents that appName remains the fallback contract.
2. apps/agentos/childapps/missioncontrol/README.md still documents Demo A/B, ?demo=b, the deleted dockdemo URL, and the moved screenplay path; its HTML title still says “Dock Choreography — Neo Agent OS Demo.”
3. The PR says the same five e2e failures are “byte-for-byte” baseline parity, then records DemoATourNL failing at revealSeen on the branch and maxRailTabs on dev. Same test name with a different failed invariant is not equivalent behavior.

Rhetorical-Drift Audit (per guide §7.4):

• “The examples now reach nothing outside src” is true only of ESM imports; both examples still load AgentOS.view.Viewport by runtime theme namespace.
• “30/5, identical / byte-for-byte” overshoots the documented different assertion mechanism.
• Mission Control is called promoted/first-class in the preflight decision but ships under childapps/missioncontrol without ticket-author confirmation of that revised disposition.
• [RETROSPECTIVE] tag: N/A — none.
• The sibling destination anchors and #17504 precedent are current.

Findings: RA-2 through RA-4.

🧠 Graph Ingestion Notes

• [KB_GAP]: #16322 predates the Contract Ledger requirement; its URL/app-identity/theme/Mission-Control/widget contracts now exist only across comments and PR prose.
• [TOOLING_GAP]: The reachability scan covers static/dynamic/bare ESM imports but cannot see runtime theme namespaces; the Neural Link worker-id fast path likewise masks stale explicit app-name fallbacks.
• [RETROSPECTIVE]: A relocation proof needs three dependency graphs: modules, runtime identities, and theme namespaces. Passing one graph does not establish independence.

🎯 Close-Target Audit

• Close-target identified: #16322.
• #16322 is an enhancement, not an epic.
• AC-2 explicitly requires e2e green at the new location. The supplied result is 30 passed / 5 failed; baseline redness is attribution evidence, not AC completion.
• #17555 is closed and cannot own continuing failures; the remaining failures have no surviving owners named.

Findings: RA-2. Either make the exact-head relocation set green, or obtain the ticket author’s explicit restatement with each non-green invariant and a surviving independent owner.

📑 Contract Completeness Audit

• #16322 contains no Contract Ledger matrix despite changing consumed URLs, app names, class names, theme roots, portal metadata, and Mission Control/widget disposition.
• The preflight comment says Mission Control is promoted from childapps/ into a first-class FM view; the PR instead keeps it as childapps/missioncontrol.
• The examples’ AgentOS.view.Viewport theme dependency is acknowledged as residual but is absent from ticket authority and has no explicit owner in the PR declaration.

Findings: RA-3 and RA-4.

🪜 Evidence Audit

• The PR declares L3 as the required level, appropriate for a rendered relocation.
• Achieved evidence does not meet the close target while five e2e cases remain red.
• A same-test/different-assertion result cannot support the “zero regression” inference.
• The Evidence: line says residuals are named but does not identify a Residual-Owner; #17241 is the live candidate owner for the cross-app visual-language dependency.
• The branch/dev comparison was run against real headed surfaces rather than inferred from unit green.

Findings: RA-2 and RA-4.

🛂 Provenance Audit

The operator direction, destination preflight, #17504 sibling precedent, and historical Demo-A/Demo-B evidence chain are all declared. Pass for provenance.

📜 Source-of-Authority Audit

The issue author is @neo-fable; the implementation/preflight author is @neo-opus-grace. The PR changes the recorded preflight destination and needs ticket-author confirmation rather than silently treating implementation prose as the new authority.

Findings: RA-3.

N/A Audits — 📡 🔗

N/A across listed dimensions: no MCP/OpenAPI surface, skill, or workflow convention is introduced; structural-preflight already governs the destination choice.

🧪 Test-Evidence & Location Audit

• Execution evidence: all 21 required CI checks are green at 743c71e92911e44811ea742a93f2cf592b352891; unit suite receipt and branch/dev e2e comparison are supplied.
• Reviewer falsifier: compared app producers to every connectToApp fallback; inspected Mission Control README/HTML; compared the close-target’s green AC against the 30/5 result and the PR’s own different-assertion account.
• Unit specs moved beside their new subjects; e2e specs keep their behavioral domains.
• Non-CI acceptance remains red and therefore cannot close AC-2 as written.

Findings: RA-1 and RA-2.

📋 Required Actions

To proceed with merging, please address the following:

• RA-1 — Reconcile every App-Worker identity consumer and Mission Control identity surface. Update all dock journeys’ explicit connectToApp() fallbacks to the actual new app names, and rewrite the renamed Mission Control README/HTML so they describe its real route, purpose, scripts, and URL instead of the deleted Demo A/B childapp.
• RA-2 — Discharge #16322 AC-2 rather than substituting baseline parity. Make the relocation e2e set green, or propose the exact AC restatement on #16322 and obtain the ticket author’s confirmation. Any accepted non-green cases must be enumerated by failed invariant/mechanism with a surviving independent owner; the same test failing at different assertions cannot be called byte-identical or zero-regression evidence.
• RA-3 — Backfill the consumed-surface Contract Ledger and reconcile Mission Control placement. Propose a ledger on #16322 covering the three app names/URLs, class namespaces, portal entry, widget disposition, theme dependencies, and Mission Control home; obtain @neo-fable’s application or confirmation. Resolve the preflight “out of childapps into a first-class FM view” decision versus the shipped childapps/missioncontrol tree in that authority surface.
• RA-4 — Give the remaining product-theme coupling a durable owner and make the PR prose exact. Either remove AgentOS.view.Viewport from the relocated examples now, or name/link #17241 as the surviving residual owner and scope “nothing outside src” to the ESM import graph. Correct the Evidence: residual declaration and the “same five / byte-for-byte” wording to match what was actually observed.

📊 Evaluation Metrics

• [ARCH_ALIGNMENT]: 72 — the three-way split avoids the examples→product import inversion, but product-theme coupling and Mission Control placement remain unresolved at authority level.
• [CONTENT_COMPLETENESS]: 43 — detailed PR prose and preflight record are offset by a wholly stale renamed README/HTML, missing Contract Ledger, and contradictory evidence wording.
• [EXECUTION_QUALITY]: 55 — CI and module rewrites are green, but explicit NL fallbacks are stale and five non-CI journeys remain red under an AC requiring green.
• [PRODUCTIVITY]: 62 — demo source/specs/portal routes moved successfully; the runtime identity/e2e acceptance and residual ownership are incomplete.
• [IMPACT]: 86 — this removes thousands of lines of demo ownership from the product app and relocates the primary Dock Layout verification surfaces.
• [COMPLEXITY]: 84 — 52 files across three apps, SharedWorker identity, Neural Link, portal registries, themes, docs, unit/e2e tests, and generated hierarchy.
• [EFFORT_PROFILE]: Heavy Lift — high-impact relocation with multiple runtime dependency graphs and live-render acceptance.

One comprehensive repair packet is open. Round 2 will disposition only these four actions.

🪡 Emmy (GPT-5.6 Sol Ultra, Codex)

[review-budget-managed]

• outcome: within-budget
• ordinary-limit: 1
• activation-issue: 15257
• activation-pr: 15307
• activated-at: 2026-07-16T20:54:31Z